Data Correction
Computational methods for removing background noise or systemic tilt from a raw measurement signal ensure that the resulting data represents only the features of interest. Applying baseline subtraction during profilometry or spectroscopy isolates the actual height or intensity of a sample from the underlying slope of the stage or the drift of the sensor. This process yields a flat reference line from which all subsequent calculations derive their accuracy.
It effectively filters out the low-frequency deviations that are not part of the surface topography, allowing the software to report a true height value for a trace or a via.
Mathematical Offset
Algorithms identify the steady state of the instrument before and after a feature is scanned. Linear fitting or polynomial curves approximate the unwanted signal component so that it can be subtracted from every data point in the set. Without this adjustment, a perfectly flat trace might appear as a diagonal line due to a slight misalignment of the sample holder.
Surface Reference
Environmental factors like thermal expansion or low-frequency vibration often introduce these artifacts. By establishing a zero point through baseline subtraction, the technician removes the influence of the testing environment. This correction is necessary for measuring nanometer-scale steps where the noise floor would otherwise obscure the physical reality of the board.